Composition-dependent thermal and ionic conduction in NASICONs: Insights from molecular dynamics simulations

The ongoing quest to optimize solid-state batteries in terms of sustainability, cost, and efficiency has shifted the focus to sodium-based battery materials. This study investigates the thermal and ionic conductivities of crystalline sodium superionic conductors of the type Na1+xZr2SixP3-xO12 (x = 0, 1.5, 1.6, 2.0, 2.4, 2.5, 2.8, 3.0) using molecular dynamics simulations. Thermal conductivities are calculated using the Green-Kubo approach at temperatures of 150, 300, 623, and 900 K. Furthermore, sodium ion diffusion constants are derived from mean-square displacements. The highest thermal conductivity is obtained for x = 0 at all temperatures. For all stoichiometries, thermal conductivity decreases with increasing temperature. Among all fractional stoichiometries, the activation barrier for ion diffusion is found to be lowest for x = 2.8. The influence of ion diffusion on thermal conductivity is shown to be negligible by artificially altering the activation barrier. However, we demonstrate that the standard Green-Kubo method strongly overestimates thermal conductivity in stoichiometries with high ionic diffusivity because of spurious convective heat flux contributions. Taking into account only the conductive part of the heat flux yields reliable results that match those from nonequilibrium simulations, cepstral analysis accounting for sodium ion mass flux, and experiments. The underlying thermal transport mechanism is analyzed using the theory of Allen and Feldman for disordered solids. The dominant transport mechanism across all investigated temperatures and stoichiometries is found to be the diffusive propagation of delocalized, non-plane-wave-like vibrational modes called diffusons. However, in the end members at 150 K, competing mechanisms are also likely to occur.